等离子体与火药相互作用过程的系统辨识

【作者】 姜久菊；

【导师】 李海元；

【作者基本信息】 南京理工大学 ， 系统工程， 2008， 硕士

【摘要】 电热化学炮的内弹道性能依赖于电能和火药化学能的有效利用率,而等离子体与火药的作用规律是影响电热化学发射中能量利用效率的关键因素。基于此情况,本文以密闭爆发器内等离子体与火药相互作用过程中的电功率和压力之间的关系以及压力和燃速之间的关系作为研究对象,并建立它们的关系辨识模型,根据这些模型可以由电功率预测出压力、再由压力预测出燃速。这些研究为等离子体与火药相互作用的机理研究提供了一种研究手段。本文的主要内容如下:(1)应用小波变换方法进行信号除噪。因为密闭爆发器实验信号中的噪声和燃烧现象造成了数据抖动,利用传统的滤波方法对所测的信号进行除噪效果不理想。而本文根据小波理论对所测的信号进行时频分解,并通过高频滤波滤掉了耦合在信号中的噪声(2)利用神经网络方法,建立了等离子体与火药相互作用过程的“电功率-压力”辨识模型。该模型经过网络训练和测试后所得的最终模型的输出结果与实测的数据误差很小,说明建立的模型可以准确地对压力信号进行预测。(3)利用小波神经网络方法结合燃速计算公式建立“压力-燃速”辨识模型。该模型仿真后得到的结果基本符合实际情况。该模型可用于压力与燃速之间规律的分析,并可以指导以后的实验。更多还原

【Abstract】 The interior ballistic performance of the electrothermal-chemical gun depends on effective utilization of the electrical energy and chemical energy. The interaction of plasma with propellant is the pivotal factor that influences energy efficiency of electrothermal chemical launch. Under this condition, the paper takes the relationship of electric power -pressure and the relationship of pressure- burning rate as researching objects, and establishes their relations identification models. The Models can predict pressure according to electric power, and then predict the burning rate. The researches provides a new research method for improving the ballistic performance.The main contents are as follows:Firstly, the knowledge of the wavelet transform is used for denoising. While measuring the signals in the closed bomb, it will make the data dithering owing to the noises and combustion phenomena. The paper does the decomposition of time-frequency signals according to the theory of wavelet and filters out high-frequency signals in the coupling of the useful signals.Secondly, the paper establishes the identification model of "electric power-pressure" in Plasma-Propellant interactions process using wavelet neural network of knowledge. After the network training and testing, the error between the final model output and the actual data is very small, which proves that the final model can accurately predict the pressure signal.Finally, the paper establishes the identification model of "pressure - the burning rate" based on the burning rate formula and exerting the knowledge of wavelet neural network. The simulated results are in accord with the actual situation basically, which proves that we can master the law of electric power-pressure and the pressure-burning rate. Moreover, the research results can guide the future researches and reduce times of the experiments.更多还原